recharts
Version:
React charts
90 lines • 4.41 kB
JavaScript
import { useCallback, useRef } from 'react';
/**
* Small state machine shared by animated components that need interruption-safe
* animations.
*
* Why this exists:
* Recharts stores the latest visible animation frame in mutable refs so the next
* animation can resume from that exact geometry. That works well, but there is a
* subtle trap: when an animation is interrupted, React may render several times
* before the new animation has actually emitted its own `animationElapsedTime=0` frame. If we keep
* reading and writing the same live ref during that window, the "start" value of
* the new animation can drift, which produces visible jumps.
*
* This hook separates those two responsibilities:
* - `startValue` is a frozen snapshot of the previous animation state, captured
* once per animation cycle and kept stable while that cycle is being matched
* and interpolated.
* - `previousValueRef.current` remains the mutable "latest visible frame" store
* that future animations can resume from.
*
* The hook does not know anything about points, baselines, sectors, or shapes.
* It only manages *when* a snapshot is captured and *when* new frames are allowed
* to overwrite the mutable ref.
*
* Lifecycle:
* 1. When `animationInput` changes by reference, a new cycle begins. We capture
* the current ref value into `startValue` and temporarily block writes.
* 2. When the new animation renders `animationElapsedTime=0`, we unlock writes. This ensures the new
* animation has had a chance to render its true starting frame before any live
* ref gets updated.
* 3. For `animationElapsedTime > 0`, callers may commit the visible frame back into the mutable ref.
* Callers can still veto that with `canCommit=false` (for example when a Line
* needs to wait until SVG path length has been measured).
* 4. At `animationElapsedTime=1`, we also refresh the frozen snapshot so subsequent rerenders in the
* completed state observe the finished geometry.
*/
export function useAnimationStartSnapshot(animationInput, previousValueRef) {
/*
* Stores the identity of the animation cycle we are currently serving.
* As soon as this changes, we know we need a brand new frozen snapshot.
*/
var previousAnimationInputRef = useRef(animationInput);
/*
* Frozen start-of-cycle value used for interpolation.
* This is the value callers should match against for the whole duration of the
* current animation, even if the live ref is updated many times afterward.
*/
var startValueRef = useRef(previousValueRef.current);
/*
* Prevents us from writing back into the live ref until the new animation has
* actually rendered its own animationElapsedTime=0 frame. This avoids "pre-start" renders from
* accidentally rebasing the next animation onto already-shifted geometry.
*/
var isReadyToCommitRef = useRef(true);
if (previousAnimationInputRef.current !== animationInput) {
// New animation cycle: capture exactly one frozen starting snapshot.
previousAnimationInputRef.current = animationInput;
startValueRef.current = previousValueRef.current;
// Writes stay blocked until the new cycle acknowledges its animationElapsedTime=0 frame.
isReadyToCommitRef.current = false;
}
var syncStepValue = useCallback(function (stepValue, animationElapsedTime) {
var canCommit = arguments.length > 2 && arguments[2] !== undefined ? arguments[2] : true;
if (animationElapsedTime === 0) {
/*
* animationElapsedTime=0 is the handshake that says: "the new animation has now rendered its
* own starting frame". We do not write anything yet; we only allow later
* in-flight frames to be committed safely.
*/
isReadyToCommitRef.current = true;
return;
}
if (animationElapsedTime === 1) {
// Keep the frozen snapshot aligned with the fully completed geometry.
startValueRef.current = stepValue;
}
if (animationElapsedTime > 0 && isReadyToCommitRef.current && canCommit) {
/*
* Commit the latest visible frame so a future interruption can resume from
* exactly what the user saw on screen most recently.
*/
// eslint-disable-next-line no-param-reassign
previousValueRef.current = stepValue;
}
}, [previousValueRef]);
return {
startValue: startValueRef.current,
syncStepValue
};
}